Braz. J. Phys. vol.36 número4A

Radio Detection of Cosmic Rays with LOPES

LOPES Collaboration

C. Grupen,1W.D. Apel,2T. Asch,3A.F.Badea,2, 4L. B¨ahren,5K. Bekk,2A. Bercuci,6M. Bertaina,7P.L. Biermann,8 J. Bl¨umer,2, 9H. Bozdog,2I.M. Brancus,6S. Buitink,10M. Br¨uggemann,1P. Buchholz,1H. Butcher,5A. Chiavassa,7 F. Cossavella,9K. Daumiller,2F. Di Pierro,7P. Doll,2R. Engel,2H. Falcke,5, 8, 10H. Gemmeke,3P.L. Ghia,11R. Glasstetter,12 A. Haungs,2D. Heck,2J.R. H¨orandel,9A. Horneffer,10T. Huege,2K.H. Kampert,12Y. Kolotaev,1O. Kr¨omer,3J. Kuijpers,10

S. Lafebre,10H.J. Mathes,2H.J. Mayer,2C. Meurer,2J. Milke,2B. Mitrica,6C. Morello,11G. Navarra,7S. Nehls,2 A. Nigl,10R. Obenland,2J. Oehlschl¨ager,2S. Ostapchenko,2, 13S. Over,1M. Petcu,6J. Petrovic,10T. Pierog,2S. Plewnia,2

H. Rebel,2_{A. Risse,}14_{M. Roth,}2_{H. Schieler,}2_{O. Sima,}6_{K. Singh,}10_{M. St¨umpert,}9_{G. Toma,}6_{G.C. Trinchero,}11 H. Ulrich,2J. van Buren,2W. Walkowiak,1A. Weindl,2J. Wochele,2J. Zabierowski,14J.A. Zensus,8and D. Zimmermann1

1_{Siegen University, Department of Physics, Emmy-Noether-Campus, Walter-Flex-Strasse 3, D-57068 Siegen, Germany} 2_{Institut f¨ur Kernphysik, Forschungszentrum Karlsruhe, 76021 Karlsruhe, Germany}

3_{Institut f¨ur Prozessdatenverarbeitung und Elektronik, Forschungszentrum Karlsruhe, 76021 Karlsruhe, Germany} 4_{on leave of absence from National Institute of Physics and Nuclear Engineering, Bucharest, Romania}

5_{ASTRON, 7990 AA Dwingeloo, The Netherlands}

6_{National Institute of Physics and Nuclear Engineering, 7690 Bucharest, Romania} 7_{Dipartimento di Fisica Generale dell’ Universita, 10125 Torino, Italy}

8_{Max-Planck-Institut f¨ur Radioastronomie, 53121 Bonn, Germany}

9_{Institut f¨ur Experimentelle Kernphysik, Universit¨at Karlsruhe, 76021 Karlsruhe, Germany} 10_{Department of Astrophysics, Radboud University, 6525 ED Nijmegen, The Netherlands}

11_{Istituto di Fisica dello Spazio Interplanetario, INAF, 10133 Torino, Italy} 12_{Fachbereich C – Physik, Universit¨at Wuppertal, 42097 Wuppertal, Germany} 13_{on leave of absence from Moscow State University, 119899 Moscow, Russia}

14_{Soltan Institute for Nuclear Studies, 90950 Lodz, Poland}

Received on 10 March, 2006

Data taken with a radio antenna array in combination with the ground-level air shower experiment KASCADE-Grande at the Forschungszentrum Karlsruhe open up the possibility to measure large extensive air showers with this new technique. The pulse height of the observed radio signals scales with the primary energy of the particles initiating the air shower. The dependence of the radio signal on the angle of the shower axis with respect to the Earth’s geomagnetic field and the coherence of the radiation suggest that the radio signal generation is due to the geosynchrotron mechanism.

Keywords: Cosmic rays; Extensive air shower; Radio emission

I. INTRODUCTION

The composition, origin, acceleration mechanism, and the propagation of the most energetic cosmic rays is still an un-solved puzzle. The solution to these problems is complicated by the fact that ultra high energy cosmic rays are extremely rare. Only one particle per square kilometer and century ar-rives at Earth with energies in excess of 1020eV. Apart from the intensity also ‘classical particles’ like protons and photons have difficulties in arriving from large distances. Photons are significantly attenuated by gamma-gamma interactions with blackbody, infrared, or optical photons, likeγ+γ→e++e−, leading to typical attenuation lengths ofλ≈10 kpc. Also en-ergetic protons (E≥6·1019eV) undergo photonuclear inter-actions with the numerous blackbody photons, likeγ+p→ ∆+_→n+π+_orγ+p→∆+_→p+π0_{, thereby losing a large} part of their energy. Typical attenuation lengths for these processes are λ≈10 Mpc. The consequence is that in the light of photons and protons (even more so for heavier nuclei)

only the local universe is visible. Neutrinos would in principle provide a way out of this dilemma. However, their detection involves considerable experimental difficulties [1].

Because of intensity reasons it is impossible to detect high energy cosmic rays (E≥1 PeV) outside the Earth’s at-mosphere. Detectors with sufficient collection power cannot be flown on satellites. Therefore an effective detection of air showers, initiated by energetic primaries, at mountain alti-tudes or ground level becomes mandatory.

shower size by integrating over the longitudinal development in the atmosphere could be combined with a near 100% duty time. In addition, the radio detection can be accomplished at moderate costs in detectors compared to sampling or fluores-cence techniques, but maybe not so low at costs in the required network.

II. GENERATION AND DETECTION OF RADIO SIGNALS

The majority of air shower particles are electrons, positrons and photons. Initially one third of the energy of the primary particle is transferred to neutral pions which feed the electro-magnetic cascade via their two-photon decay. Further elec-tromagnetic energy results fromπ0_{production in interactions} of secondary charged particles (π±_,K±_{, . . .), so that}

eventu-ally a large fraction of the primary energy is observed in elec-trons and posielec-trons. These particles travel with approximately the speed of light through the atmosphere. The thickness of the shower disk moving through the atmosphere is just a few meters (corresponding to≈10 ns), so that coherent phenom-ena are expected for radio frequencies up to a few 100 MHz. Since the velocity of the electrons and positrons exceeds the velocity of light in air (velectron≥c/n, wherenis the index of refraction of air), one would expect a coherent production of Cherenkov radiation due to the Askaryan effect [6]. Askaryan argued that in electromagnetic showers an excess of electrons over positrons must develop, because on the one hand positron annihilation removes part of the positive charge while on the other hand Compton andδelectrons add new negative charge to the shower. For the generation of Cherenkov radiation the effect of negative and positive charges – due to their charge imbalance – would no longer cancel, giving rise to a net Cherenkov signal. This phenomenon is expected to become dominant in dense media, like water or ice. On the other hand the electrons and positrons are deflected in the Earth’s magnetic field providing a source of geosynchrotron radia-tion, which is believed to be the dominant mechanism of radio generation in the atmosphere. In air this process produces a stronger emission than the Askaryan effect, since it involves all of the charged particles, rather than just the excess charge as in the coherent Cherenkov emission [7]. Other genera-tion mechanisms or interpretagenera-tions like dipole radiagenera-tion, due to charge separation in the Earth’s magnetic field [8], or the effect of molecular-field bremsstrahlung [7] are believed to be of minor importance.

The advantage of radio detection over air shower sampling or air fluorescence detection can be characterised as follows:

• simple, robust, and inexpensive detectors;

• full day operation possible (maybe except thunder-storms);

• integration over the whole air shower development, pro-viding information on the total shower energy;

• wide field of view;

• low attenuation of the radio signal.

The determination of the chemical composition of primary cosmic rays using radio detection is more difficult compared to fluorecence techniques, because the radio signal integrates over the whole shower development, while the fluorescence signal allows also to determine the position of the maximum of the shower developmentXmax, which is sensitive to the na-ture of the particle that initiates the shower.

Potential problems of the radio technique also consist in ra-dio frequency interference, which, however, can be overcome by digital filtering techniques and beam forming [9–11]. Due to the omnipresent radio frequency noise radio detection is only feasible at high energies (≥1016_{eV) where the} signal-to-noise ratio becomes favourable.

III. RADIO DETECTION AND THE LOPES ARRAY

The phenomenon of radio emission from extensive air showers was experimentally discovered in 1965 by Jelley et al. [12]. Early measurements in the sixties were compiled by Allan [13]. Theoretical descriptions of the phenomenon were provided by Kahn and Lerche [8] and Colgate [14] soon af-ter. Many activities developed in the late sixties and early seventies, but due to problems with radio interference, poor time resolution, low statistics, and limited angular acceptance this field of research soon quieted down and became dormant. Only with the availability of modern electronics and the pos-sibility of digital signal filtering and beam forming [9–11, 15] the radio detection of extensive air showers became rejuve-nated as a realistic alternative to classical air shower detection. In addition, more comprehensive theoretical descriptions and Monte Carlo codes for the understanding of the phenomenon pushed the field [7, 16–19].

FIG. 1: Inverted V dipole antenna for the LOPES experiment inte-grated into the KASCADE array. Only two ‘legs’ of the dipole are equipped with antennas, such that only one polarisation in east-west direction can be measured. In the background one can see some scin-tillator sampling detectors in wooden housings on the campus of the Forschungszentrum Kalrsruhe.

Netherlands predominantly meant for radio astronomy. Ini-tially LOPES comprised 10 simple inverted V dipole antennas (figure 1), called LOPES 10. It was later extended to 30 anten-nas (LOPES 30). These antenanten-nas are installed on the site of the original and extended Karlsruhe air shower array (KAS-CADE [2] (figure 2) and KAS(KAS-CADE-Grande [21]) at the Forschungszentrum Karlsruhe. KASCADE and KASCADE-Grande represent multi-detector setups which allow high per-formance estimations of charged particles (electrons, photons, muons, and hadrons) in extensive air showers in the primary energy range from 1014to 1018eV.

200 m

13 m

Central Detector

Array Cluster

Detector Station

Electronic Station Muon Tracking Detector

Grande Station Radio Antenna (LOPES-10)

FIG. 2: The main detector components of the KASCADE experiment consisting of 16 clusters of scintillation counters of the field array (in total 252 detector stations), a special muon tracking detector, and a central detector incorporating a large hadron calorimeter.

The 10, respectively 30 LOPES dipole antennas are placed essentially inside the original KASCADE array (200× 200 m2) with some outside in the KASCADE-Grande detec-tor system (800×800 m2, Fig. 3). The electron and muon components of the air showers are measured in small stations containing several scintillation counters each (Fig. 4).

The antennas of the LOPES detector system operate in the frequency range 40–80 MHz. They are aligned in east-west direction and are therefore only sensitive to the east-west lin-ear polarisation component of the radio emission. The readout of the LOPES antennas is triggered by the KASCADE array. The readout window for the antennas is 0.8 ms wide, and it is centered on the KASCADE trigger. The KASCADE array provides the information on the energy of the primary parti-cle initiating the air shower, the direction of the shower axis, and the distance of the shower axis from the antennas. These data are derived from the lateral distribution of electrons, pho-tons, and muons as observed in the various detectors of the

KASCADE

LOPES−10

Grande stations

X [m]

Y [m]

FIG. 3: Sketch of the Grande experiment. KASCADE-Grande is an extension of the smaller KASCADE array. It has 37 additional detector stations. The position of the first 10 LOPES an-tennas is indicated.

240 cm light-collector

glas fiber cable HV, anode and

dynode connectors

photo-multiplier

argon

10 cm lead 4 cm iron

e/γ - detector

5 cm liquid scintillator

µ - detector

3 cm plastic scintillator

FIG. 4: Cross section of one electron and muon detection unit con-sisting of scintillation counters in the stations of the KASCADE ar-ray.

KASCADE and KASCADE-Grande arrays. The shower size, or the energy of the primary particle, initiating the shower, is usually obtained from the electron component in the air shower, or by a combination of the electron and muon size.

amplitude of the radio signalεν per unit bandwidth induced

by the air shower can be described by the following formula,

εν=20·₁₀17E0_eV·sinα·cosθ·exp(− R R0(ν,θ)) [

µV m·MHz], where

E0– primary energy of the shower in eV,

α– angle between the shower axis and the geomagnetic field,

θ– zenith angle of the shower,

R– distance to the shower axis, and

R0(ν,θ) =110 m at 55 MHz is a scaling radius. This formula, obtained from measurements of small band-widths, should be considered as a guideline for radio detec-tion. In particular, the numerical scaling factor of 20 has to be considered with caution.

LOPES data have been analysed in many different ways. One of the main aims of the pioneering LOPES experiment was to establish the feasability of radio detection of air show-ers, and to clarify the production mechanism of radio emis-sion. In special analyses the behaviour of distant events [10], the question of the zenith-angle dependence of radio produc-tion, and the possibility for neutrino detection based on radio emission [22] was investigated.

In general, LOPES data are selected by the following cuts:

• trigger from the KASCADE-Grande arrays;

• a successful reconstruction of the air shower with the core position inside the KASCADE-Grande arrays;

• zenith angle smaller than 50◦;

• E0≥4·1016eV;

• distance cut depending on the energy (e.g.R≤200 m at 1017eV,R≤400 m at 4·1017eV);

• approximately uniform pulse height in all antennas.

The analysis in this paper is based on LOPES 10 data. The LOPES 10 antennas are essentially placed inside the origi-nal, smaller KASCADE array (see figure 3). Therefore, the analysis relies predominantly on properties given by the orig-inal KASCADE set-up. The offline analysis of the LOPES 10 data required a trigger from the traditional, smaller sampling array, and the following cuts were applied [9]:

• the KASCADE event readout was ‘good’, in the sense that the array processor was working properly;

• the distance of the shower core to the KASCADE array centre was less than 91 m, corresponding to the effective radius of the KASCADE array;

• the electron number was greater than 5·106_,orthe trun-cated muon number (i.e. the number of muons in the distance range of 40–200 m from the shower axis) was greater than 2·105.

The last criterion ensures that mainly large events are se-lected, where a significant radio signal is expected.

The analysis of the data is done offline and proceeds along the following lines:

• correction for instrumental delays;

• gain correction for all electronic components;

• filtering of narrow band radio frequency interference;

• selection of antennas with significant signals;

• digital beam forming in the direction of the shower [9– 11];

• quantification of the peak parameters.

The digital beam forming presents an essential step in the analysis. The signals from the antennas are shifted and cross-correlated in such a way that the coherent pulse from the air shower is optimised using the information of the zenith angle and azimuthal angle from the KASCADE arrays [9]. Since the bulk of the radio signal originates mainly from the large number of electrons in the shower maximum – i.e. from an extended point source – it is also expected that the radio wave-front, instead of being a plane wave, should have some curva-ture. During the reconstruction procedure this radius of cur-vature is taken into account by iterating it as a free parameter until the radio beam is maximal [10, 11]. This curvature para-meter relates essentially to the ‘electromagnetic’ shower front. Typical radio events have amplitudesεν in the frequency

range of 40–80 MHz of someµV/(m·MHz) corresponding to electrical fields ofE≈150µV/m. It has to be noted that at coherence the radio powerPis proportional to the square of the electric field.

IV. RESULTS

The big advantage of radio detection of air showers in a phased array is that radio antennas in principle look in all directions in the sky at the same time. For a known angle of incidence of the air shower – in this case provided by the KASCADE arrays – the signals can be phased in that direc-tion by remotely adjustable delays. Offline many different di-rections can be tested by the method of phased digital beam forming [10, 11]. This technique also works without using the information of the shower direction.

The electric field as measured at each antenna corrected for the arrival direction still shows a lot of incoherent noise (see figure 5, [23]).

FIG. 5: Superposition of the electric fields of all antennas for an individual shower event. The various pulses are delay-corrected for the arrival direction of the air shower [23].

event as in figure 5 is shown in figure 6 [23]. The time offset of≈1.8µs of the radio signal with respect to the KASCADE trigger is due to hard-wired delays.

Figure 7 shows an example of a false colour radio map of an air shower event. By analyzing the arrival times of the radio flashes the direction of incidence of the shower can be independently determined. The result of such a shower recon-struction is seen as central blob in the figure. Other weak sig-nals surrounding the central brightness maximum result from interferometer sidelobes by the sparse radio array and from background noise in the radio signals [15].

FIG. 6: Sum of delay-corrected electric fields from all antennas squared for an individual event [23].

The pulse height of the radio signal is proportional to the integrated track length of all electrons and positrons in the air shower. It should depend simultaneously on a number of other parameters, namely

• number of muons;

FIG. 7: False colour radio map of an air shower event. The recon-structed image of the shower is seen as bright blob at the centre of the figure. Other weak signals surrounding the central brightness max-imum result from interferometer sidelobes by the sparse radio array and from background noise in the radio signals [15]. AZEL stands for Azimuth and Elevation.

• distance to the shower axis;

• angle to the geomagnetic field;

• azimuthal and zenith angle of the shower.

Since electrons and positrons are responsible for the radio-synchrotron signal generation, the radio signal should only depend on the total number of electrons and positrons (the ‘electron number’) in the shower. However – due to propa-gation effects in the atmosphere and the way of measurement – the recorded radio signal also depends on the azimuthal an-gle, the zenith anan-gle, and the fact that only one polarisation is measured. In particular, due to the generation mechanism, the dependence of the radio yield on the angle of the shower axis with respect to the geomagnetic field plays a special role.

It is not easy to disentangle the various correlations for these dependencies. When, for example, the dependence of the radio pulse height on the geomagnetic angle is to be ex-tracted, the dependence of the radio signal on the shower size and the lateral distance of the shower must be considered and corrected for. In the same way, the dependence of the radio signal on the electron or muon number must take into ac-count that geomagnetic angle and core distance effects have to be corrected for. In the following it is tried to present var-ious functional dependencies of two parameters each where the correlation with other parameters is taken into account.

LOPES – as derived from measurements in coincidence with the KASCADE-Grande array [10] – depends on the primary energy. It is appoximately 60% for primaries with energies exceeding 2·1017eV. There is a number of reasons that this efficiency cannot be 100%. Firstly, there is a detection thresh-old for radio showers which, among others, depends on the ambient radio background. Secondly, the present LOPES de-tection system is sensitive to only one polarization direction, and thirdly, if the shower is incident along the direction of the geomagnetic field, there will be very little synchrotron emis-sion.

The following results are based on a sample of events that have been collected from January to September 2004 [9]. Out of 630 000 events coincident with KASCADE 556 626 ‘good’ data from KASCADE have been found. Since the ef-ficiency for radio detection is strongly dependent on the pri-mary energy, only high energy showers have been selected. For the subsample of energetic showers 412 ‘good’ KAS-CADE events were selected, out of which 228 had a clear radio peak, corresponding to an efficiency for this sample of 55%. For these LOPES 10 events the radio pulse height nor-malised to the muon number (i.e. essentially to the shower size or equivalent primary energy determined from the muon data) and distance to the shower axis shows a clear correlation with the geomagnetic angle (figure 8).

FIG. 8: Radio pulse height normalised to the muon number and dis-tance to the shower axis as a function of the geomagnetic angle in terms of 1−cosα[9].

Primaries incident parallel to the geomagnetic field (cosα=1) can produce very little synchrotron radiation be-cause the transverse momenta of shower particles are usually very small.

If the muon data are corrected for the geomagnetic angle effect a clear correlation of the radio pulse height with the shower size in terms of the muon number is observed (fig-ure 9).

A correlation of the radio signal with the electron number in the shower is also observed, but it is less pronounced com-pared to the muon case (figure 10).

This may be related to the following: Even though the ra-dio signal originates from all phases of the longitudinal de-velopment, and considering the fact that the signals are

atten-FIG. 9: Variation of the radio pulse height with the muon size of the shower after the signal amplitude has been corrected for the geomag-netic angle and the distance to the shower core [9]. The experimental point in the first bin is biased by the efficiency for a poor signal-to-noise ratio at low shower sizes.

FIG. 10: Variation of the radio pulse height with the electron size of the shower after the signal amplitude has been corrected for the geomagnetic angle and the distance to the shower core [9].

have an impact on the correlation of the radio signal with the electron and muon number.

FIG. 11: Variation of the radio pulse height with the primary particle energy of the shower as derived from the electron and muon size after the signal amplitude has been corrected for the geomagnetic angle and distance to the shower core [9].

Finally, figure 11 shows the dependence of the radio pulse height on the primary particle energy, which has been derived from the combined electron and muon number of the shower.

The analysis of the available statistics of radio signals for KASCADE and KASCADE-Grande triggered and selected events leads to the following results:

• angular resolution of the shower axis of better than 1◦;

• width of the radio cone of a few degrees;

• duration of the radio flash of≤45 ns.

For completeness the dependence of the radio pulse height on the distance from the shower axis is presented in figure 12.

FIG. 12: Variation of the normalised radio pulse height with distance from the shower axis after scaling for geomagnetic angle and shower size effects [9].

The radio pulse is coherent (the thickness of the shower disk is smaller than the wavelength of the emitted radio wave), the

pulse height scales like a power law with a spectral index of 1 with the shower energy, and it is correlated with the geo-magnetic angle, which is a clear indication that the production mechanism is geosynchrotron radiation.

The discussion of the zenith-angle dependence of the radio signal is somewhat intricate. For the energy range in ques-tion the total number of electrons depends very little on the zenith angle, because the shower maximum is relatively high in the atmosphere. Therefore the radio signal should also not depend significantly on the zenith angle. On the other hand, the recorded shower size at sea level does depend on the zenith angle, since the absorption of the electrons is much stronger than the attenuation of the radio signal. This leads to a smaller recorded shower size, producing less noise for the radio detection, making the radio signal look cleaner, because of the improved signal-to-noise ratio. Also, for a fixed trigger threshold, inclined showers are on average of higher energy, and since they are ‘old’, the ‘footprint’ of the radio signal is smeared over a larger area, thereby easing radio detection in a widely spaced array [19]. In addition, for larger zenith an-gles the angle with respect to the geomagnetic field becomes more favorable. If all these interdependencies are considered, it is believed that the radio pulse height should show no clear dependence neither on the zenith angle nor on the azimuthal angle if the radio signal is scaled with the muon number and the effect of the geomagnetic field is taken into account. This behaviour is also confirmed by the data [9].

V. SUMMARY AND CONCLUSIONS

LOPES has verified the phenomenon of geosynchrotron emission of extensive air showers. With digital filtering and cross-correlated beam forming the radio pulses can be mea-sured even in a radio loud environment. The radio pulse height depends – as expected – on the geomagnetic angle. After ge-omagnetic angle correction the radio pulse height is strongly correlated with the muon number, and also with the electron number, even though the correlation with the muon number is somewhat more pronounced. This means that the radio pulse height scales with the shower energy. In principle, the radio measurements also allow to determine the shower core posi-tion, the angle of emission and the shower width. Still there is room for improvement towards a detailed understanding of the interplay of the different shower parameters.

One must, however, remember that the LOPES results have so far been obtained in an offline analysis based on selected air shower data triggered by a classical air shower sampling array. A stand-alone radio detection array, like LOFAR [20], still needs to verify the results obtained so far.

VI. ACKNOWLEDGEMENTS

The authors of the LOPES Collaboration would like to thank the engineering and technical staff of the involved insti-tutes. The corresponding author (C.G.) expresses his thanks to the organisers of the XXVI Brasilian National Meeting on

Particles and Fields (BNMPF), in particular to Fernando Mar-roquim Leao de Almeida Jr., for the relaxed atmosphere at S˜ao Lourenc¸o and the hospitality extended to him. The corre-sponding author also acknowledges the careful reading of the manuscript by Tilo Stroh.

[1] C. Grupen, Astroparticle Physics, Springer, Heidelberg, 2005 [2] T. Antoni et al., KASCADE Collaboration, Nucl. Instr. Meth.

A 513, 429 (2003)

[3] M. Takeda et al., Astropart. Phys.19, 447 (2003)

[4] J.N. Matthews et al., Proc. Int. Cosmic Ray Conf. 2001, and http://hires.physics.utah.edu/

[5] Auger Collaboration, http://www.auger.org/ [6] G. A. Askaryan, Soviet Physics JETP14, 441 (1962)

[7] P.W. Gorham, SLAC Summer Institute on Particle Physics (SSI04), Aug. 2–13 (2004)

[8] F.D. Kahn, I. Lerche, Royal Society of London, Proceedings SeriesA 289, 206 (1966)

[9] A. Horneffer, Ph.D. thesis, Measuring Radio Emission from Cosmic Ray Air Showers with a Digital Radio Telescope, Max-Planck-Institut f¨ur Radioastronomie Bonn, and Univer-sit¨at Bonn, Germany (2006)

[10] W.D. Apel et al., LOPES Collaboration, submitted to Astropart. Phys. 2006

[11] A.F. Badea et al., Forschungszentrum Karlsruhe,

Wis-senschaftliche Berichte, FZKA 7229, April 2006 [12] J.V. Jelley et al., Nature205, 237 (1965)

[13] H.R. Allan, Progr. in Element. Part. and Cosmic Ray Physics, Vol. 10, 171 (1971)

[14] S.A. Colgate, J. Geophys. Res.72, 4869 (1967)

[15] H. Falcke et al. – LOPES Collaboration, Nature435, 313 (2005) [16] H. Falcke, P.W. Gorham, Astropart. Phys.19, 477 (2003) [17] T. Huege, H. Falcke, Astron. and Astroph.412, 19 (2003) [18] T. Huege, H. Falcke, Astron. and Astroph.430, 779 (2005) [19] T. Huege, H. Falcke, Astropart. Phys.24, 116 (2005) [20] LOFAR Collaboration, http://www.lofar.org/

[21] G. Navarra et al., KASCADE-Grande Collaboration, Nucl. In-str. Meth. A513, 429 (2003)

[22] J. Petrovic et al., LOPES Collaboration, submitted to Astron-omy and Astrophysics, June 2006